Distribution of HIV-1 infection in different T lymphocyte subsets: antiretroviral therapy-naïve vs. experienced patients

Distribution of HIV-1 Infection in Different T LymphocyteSubsets: Antiretroviral Therapy-Naıve

vs. Experienced Patients

Raul Perez,1 Sonia Gibson,1 Pablo Lopez,2 Ellen Koenig,3 Marisol De Castro,3 and Yasuhiro Yamamura2

Abstract

Memory CD4 T cells are the primary targets of HIV-1 infection, which then subsequently spreads to other Tlymphocyte subsets. Antiretroviral therapy (ART) alters the pattern of HIV-1 distribution. Blood samples werecollected from ART-naıve or -experienced HIV-1 patients, and the memory and naıve subsets of CD4þ and CD8þ

T lymphocytes, respectively, were isolated by cell sorting. DNA was extracted and the HIV-1 env C2/V3 regionPCR amplified. Amplicons were cloned and sequenced, and genetic relatedness among different HIV-1 com-partments was determined by the phylogenetic analysis of clonal sequences. The viral V3 sequence of HIV-1 ineach compartment was analyzed by using webPSSM to determine CCR5 or CXCR4 coreceptor binding propertyof the virus. The direction of viral migration among involved compartments was determined by using theMacClade program. In ART-naıve patients, HIV-1 was generally confined to the memory CD4 T (mT4) cellcompartment, even though in a few cases, naıve CD4 T (nT4) cells were also infected. When this occurred, theHIV-1 gene migrated from nT4 to mT4. In contrast, HIV-1 was detected in nT4 and mT4 as well as in thememory CD8 T (mT8) compartments of ART-experienced patients. However, no clear pattern of directionalHIV-1 gene flow among the compartments could be determined because of the small sample size. All HIV-1–infected T cell compartments housed the virus that used either CCR5 or CXCR4 as the coreceptor.

Introduction

HIV-1 infection is generally perceived as systemic,where the transmitted virus spreads to various lym-

phoid organs, while rapidly mutating and developing quasi-species.1,2 However, there also is an emerging awareness thatHIV-1 infections become highly compartmentalized in thebody of a patient, with each lymphoid compartment seededby a distinct viral lineage. For example, HIV-1 in the centralnervous system has long been recognized to be distinct fromthe virus in the circulating blood of the same patient.3–5 HIV-1in the male or female genital secretion has also been shown tobe distinct from that in the blood of the same individuals.6–9

While memory CD4 T (mT4) cells are the primary targets ofthe infection,10–13 HIV-1 is also detected in other T cell subsets,including naıve CD4 T (nT4) and, in some cases, memory CD8T (mT8) cells in patients with extensive antiretroviral therapy(ART) histories.14–16 Involvement of double-negative (i.e.,CD4-CD8-) T cells has also been described.17 Previous studieswith ART-experienced patients have shown that each T cell

subset was infected by a mutually distinct lineage of the vi-rus.15,16 In the present study, we further examined if HIV-1infection of multiple T cell subsets is the phenomenon asso-ciated only with ART-experienced patients. While there arereports of CD8-tropic HIV-1 that specifically infects matureCD8 T cells,18,19 HIV-1 in mT8 of ART-experienced patientswas described to be a CD4-tropic virus that infected theCD3þCD4þCD8þ ‘‘immature’’ or ‘‘pro-thymic’’ T cells, whichthen ‘‘phenotypically’’ matured under thymic influences, be-coming CD4-CD8þ T cells.16 We investigated the directionalflows of the virus among different T cell subsets of each pa-tient whenever multiple T cell subsets of a patient were HIV-1infected.

Materials and Methods

Human research materials and isolation of differentT cell subsets

The study utilized, without any personal identifiers, un-used portions of venous blood samples of HIV patients that

1Department of Internal Medicine, Wayne State University, Detroit, Michigan.2Ponce School of Medicine AIDS Research Program, Ponce, Puerto Rico.3Instituto Dominicano de Estudios Virologicos, Santo Domingo, Dominican Republic.

AIDS RESEARCH AND HUMAN RETROVIRUSESVolume 27, Number 4, 2011ª Mary Ann Liebert, Inc.DOI: 10.1089/aid.2010.0176

399

had been sent to our reference laboratory for CD4T cell enu-meration and viral load assessment. The protocol was there-fore classified, following an expedited review, as ‘‘exempt’’from further review requirements by the Ponce School ofMedicine Institutional Review Board, FWA00000345. All testsamples were obtained as EDTA-treated blood and were se-lected for inclusion in the study by convenience (e.g., sufficientblood volumes, medium-range CD4 T cell counts and no ex-tensive hemolysis, etc.). Peripheral blood mononuclear cells(PBMC) were separated from blood by centrifugation in Ac-cutubes containing Histopaq-1077 (Sigma-Aldrich, St. Louis,MO), according to the manufacturer’s recommended proto-col. Plasma was separated and stored for testing. The cellswere washed twice with RPMI-1640 (Sigma-Aldrich) sup-plemented with10% fetal bovine serum (FCS) (Caisson, SugarCity, ID), and then once more with Dulbecco’s phosphatebuffered solution (PBS) (Sigma-Aldrich) containing 1% sodi-um-azide (Sigma-Aldrich) and 2% FCS. Cells were then sus-pended in 20 ml each of anti-CD3 PE-Cy7, -CD4 PerCP, -CD8APC-Cy7, -CD45RA-FITC, and -CD45RO APC murinemonoclonal antibody-conjugates (BD, San Jose, CA). Afterincubation for 20 min at room temperature in the dark, un-bound antibodies were removed by washing twice with PBSplus 2% FCS. Cells were analyzed for each marker expressionusing a FACSAria flow cytometer and FACSDiva software(BD) and CD3þCD4þCD45RAþ(naıve CD4-T: nT4), CD3þ

CD4þCD45ROþ (memory CD4-T: mT4), CD3þCD8þCD45RAþ

(naıve CD8-T: nT8), and CD3þCD8þCD45ROþ(memory CD8-T:mT8) cells were simultaneously isolated by cell sorting. An ap-propriate isotype control was used for each conjugate to identifyaccurately the cells that were positive for each marker. Purity ofeach sorted cellular population was >99% in all cell prepara-tions. Any samples that failed to yield 1.0 x 104 cells for morethan one T cell subset were excluded from the analysis. Theminimum number of cells required was empirically set to ensurea successful PCR amplification and cloning, which are describedbelow.

DNA extraction and PCR amplification of HIV-1env C2-V3 gene

Each purified T cell subset was counted and washed andDNA was extracted and purified using a QIAamp DNABlood minikit (QIAGEN, Germantown, MD). HIV-1 RNAwas similarly purified from the matching plasma by using theQIAGEN QIAamp RNA Blood minikit. Purified RNA (25ml),serially diluted,20,21 was RT-PCR-amplified for HIV-1 env C2–V3 gene sequences. Briefly, HIV-1 C2–V3 genes were RT-PCRamplified using Roche Titan One-Tube RT-PCR kits and ABIGene Amp-9700 cyclers. The RT-PCR conditions were: reversetranscription at 608C for 10 min, inactivation at 958C for10 min, followed by 35 cycles of (948C for 30s, 608C for 30s and728 for 1:30 min) with the final extension at 728C for 7 min.The first round primer pair (50 pmole each) used was:ED31F: 5’-CCTCAGTCATTACACAGGCCTGTCCAAAG-3’and ED12R: 5’-AGTGCTTCCTGCTGCTCCCAAGAACCCAAG-3. The first round amplicons were re-amplified using thesecondary primer pair (50 pmole); C2V3F: 5’-CTGTTAAATGGCAGTCTAGC-3 and C2V3R: 5’-TGATGGGAGGGGCATACATT-3’.

Amplicons were visualized by UV in 1.5% agarose gelelectrophoresis using ethidium bromide. The resulting am-

plicons represented the HIV-1 gene nucleotide sequence po-sitions 7002 through 7217. (PCR amplification from the HIV-1proviral DNA was similarly performed but without the initialRT step). Confirmed amplicons were further processed forcloning and sequencing.

Cloning and sequencing of C2V3 amplicons

Each confirmed amplicon was cloned using TOPO TACloning Kit PCR 2.1/4.0 Topo Vectors with One Shot Top 10chemically competent E. coli (Invitrogen, Carlsbad, CA).Preparation and procedure were carried out according to themanufacturer’s recommended protocol. Approximately 15colonies were randomly picked from the plates of the lowesttemplate concentrations that successfully yielded the ampli-con; another 15 came from the plates that used x100 highertemplate concentration. The insert verification was performedby colony PCR for the C2–V3 sequence. A Fast Plasmid� Mini(Eppendorf, Westbury, NY) and a QIAprep� Spin miniprepkit (QIAGEN) were interchangeably used for the purificationof the specific plasmids, according to the respective manu-facturers’ recommended protocols. The plasmid preparationswere purified by the PCR purification EXO-SAP reaction,consisting of a mixture of E. coli exonuclease I (Epicentre�,Madison, WI) and shrimp alkaline phosphatase (Roche Mo-lecular System, Indianapolis, IN) followed by treatment witha Big Dye� Terminator� v3.1 Cycle sequencing Kit (AppliedBiosystems, Foster City, CA). Sequencing reaction was carriedout using 3.2 pmol M13 primers (Integrated DNA Technolo-gies, Coralville, IA) M13F 5’-GTAAAACGACGGCCAG-3’and M13R 5’-CAGGAAACAGCTATGAC-3’. DNA sequenc-ing was performed using an ABI 3730 automated DNA se-quencer with 48 capillaries.

Phylogenetic analysis and HIV-1 gene flowbetween T cell subcompartments

HIV-1 C2-V3 sequences were imported to SeqScape� (v2.5)software (Applied Biosystems) and aligned by using theHXB2 sequence as the reference.22 Aligned sequences wereexported in the FASTA format to a BioEdit Sequence Align-ment Editor (v7.0.5.2) (http://www.ncsu.edu/BioEdit). Todetermine the coreceptor tropism of HIV-1 in each compart-ment, we used Web PSSM, an online program (http://indra.mullins.microbiol.washington.edu/webpssm/) that is basedon the method described by Jensen et al.23

Phylogenetic relationships among different HIV-1 com-partments were analyzed by using a neighbor-joining algo-rithm, and the intra- and inter-compartmental genetic distanceswere calculated by the genetic distance program of MEGA (v3.1) software. The Tamura–Nei nucleotide substitutions modelwas used to calculate evolutionary distances. Compartmenta-lization of HIV-1 in each T cell subset was tested by the Stu-dent’s t-test, as previously described by Philpott et al.,24 andp< 0.05 was used as the indication of HIV-1 compartmentali-zation. The tree was constructed, after 10,000 bootstrapping,with the pairwise deletions and gamma distribution options.The Bayesian analysis (posterior probabilities distribution oftrees) needed for gene flow analysis was created by usingthe BEAST package of software (v.1.5.4) (http://beast-mcmc.googlecode.com/files/BEAST%20v1.5.4), according to theMonte Carlo Markov Chain (MCMC) method.25,26 The nexusformat necessary to perform the analysis was obtained using

400 PEREZ ET AL.

Readseq, a biosequence conversion tool, which was providedby European Bioinformatics Institutes (EBI) (http://www.ebi.ac.uk/cgi-bin/readseq.cgi). Consensus trees were ob-tained after repeating sufficient generations of MCMC analysis,with sampling every 1000, to obtain the values of >100 for allparameters of effective sample size.27–30 The trees were rootedwith 2 group-M C-clade (U46016, U52953), 2 group-O (L20571,L20587) and HXB2 C2–V3 sequences from GenBank. Finally, asingle ‘‘target’’ tree was obtained using TreeAnnotator (v.1.5.4).HIV-1 C2-V3 gene flow between different compartments waspair-wise determined by assessing the direction of HIV-1 viralgenetic evolution using the MacClade (v 4.08; Sinauer Associ-ates, Sunderland, MA) program, as previously described byothers.31,32 Directional gene flow was traced with the stateschanges and stasis tool and expressed by the number of calcu-lated migration events (medians) between the compartments.

Nucleotide sequence accession numbers

All the sequences used in this study have been deposited inGenBank under accession numbers HQ115756 through116168.

Results

T cell subset counts and viral loadsof the samples utilized

Twenty-one and twelve blood samples from therapy-experienced and -naıve patients, respectively, met ourscreening criteria of having a sufficient number of each T cellsubset and yielding HIV-1 env C2V3 PCR amplicons for one ormore of T cell subsets (Table 1). For each sample, the numberof cells obtained from each T cell subset (by cell sorting) and

Table 1. Numbers of Subset T Cells Recovered by Cell Sorting

ART-experienced

CD4 count ID # CD4RA1 CD4RO1 CD8RA1 CD8RO1 Viral load

< 200 85 3 N/A2 2.2 E43(R5) 5.7 E4 5.0 E4 4,05055 5 N/A 1.9 E4 (R5) 1.6 E5 6.1 E4 58,328

198 10 7.2 E4 (R5) 6.8 E4 (R5) 2.7 E5 3.7 E5 111,22392 13 2.0 E4 1.8 E4 2.5 E5 (X4) 7.8 E4 27,094

131 15 N/A 7.8 E4 (R5) 9.9 E4 1.4 E4 164,893193 26 4.6 E3 8.2 E4 (X4) 1.4 E5 8.1 E4 <48

200* 500 440 6 6.8 E4 (R5) 8.9 E4 (R5) 7.4 E4 9.7 E4 ND4

283 7 5.1 E3 1.6 E5 (R5) 7.5 E4 9.7 E4 7,848378 21 3.6 E4 (R5) 1.5 E5 (R5) 4.5 E4 9.3 E4 195362 24 3.4 E4 7.3 E4 (R5) N/A 1.0 E5 10,648427 25 1.8 E5 1.6 E5 (R5) 1.9 E5 6.4 E4 <48478 27 1.8 E5 2.0 E5 (X4) 3.4 E5 2.5 E5 (X4) <48

> 500 907 19 6.8 E4 8.7 E4 (R5) 1.1 E5 8.2 E4 5,650752 20 1.2 E5 3.0 E5 (R5) 2.1 E5 4.5 E4 302609 22 2.6 E4 1.9 E5 (R5) 2.7 E5 6.7 E4 (R5) <48605 16 3.3 E4 1.5 E5 (R5) 8.6 E4 8.8 E5 (R5) ND660 17 1.3 E5 3.2 E5 (R5) 2.2 E5 1.6 E5 ND

1,077 14 2.3 E6 (X4) 3.6 E5 (X4) 6.5 E5 2.1 E4 ND669 18 N/A 1.6 E5 2.8 E5 1.8 E5 (R5) <48594 30 1.1 E5 (R5) 2.2 E5 (R5) 4.2 E5 1.9 E5 ND

1,094 31 1.6 E5 (X4) 2.6 E5 (R5) 1.8E5 1.9 E5 (X4) 76Total 2 X4/4 R5 3 X4/16 R5 1 X4/0 R5 2 X4/3 R5

ART-naıve

Sample ID CD4 counts CD4RA CD4RO CD8RA CD8RO Viral load

DR26 408 4.3 E4 9.5 E4 (R5) 1.1 E5 1.1 E5 NADR27 331 3.2 E4 6.5 E4 (R5) 2.2 E5 2.0 E5 NADR06 1,022 2.4 E4 6.5 E4 (R5) 4.3 E4 8.4 E4 NADR20 326 3.8 E4 1.8 E5 (R5) 4.7 E5 1.7 E5 21,800DR09 547 2.7 E5 3.6 E5 (R5) 2.5 E5 2.5 E5 <50DR19 720 1.7 E5 3.5 E5 (R5) 7.9 E5 3.9 E5 NADR17 618 1.2 E5 2.1 E5 (R5) 2.0 E5 1.2 E5 NADR18 460 1.9 E5 (X4) 2.6 E5 (X4) 6.6 E5 1.1 E6 5,660DR22 512 2.7 E4 5.5 E4 (R5) 1.8 E5 7.7 E4 NADR23 488 1.4 E5 2.4 E5 (R5) 2.4 E5 6.6 E5 6,780DR25 399 7.1 E4 2.3 E5 (R5) 4.2 E5 3.4 E5 NADR21 447 2.4 E4 (X4) 3.4 E4 (X4) 5.2 E4 6.3 E4 NATotal 2 X4/0 R5 2 X4/10 R5

1CD4 or CD8 T cells were further separated by cell-sorting as CD45RA (naıve) and CD45RO (memory) subsets. Each subset was expressedas CD4RA, CD4RO, CD8RA, and CD8RO.

2N.A., not available; a sufficient number of cells could not be recovered after cell-sorting.3The 2.2 E4 designation stands for 2.2�104.

HIV-1 DISTRIBUTION IN T LYMPHOCYTE SUBSETS 401

the plasma viral load are provided. In total, 31/79 isolated Tcell compartments used for PCR amplification were positivefor HIV-1 proviral C2–V3 gene. As anticipated in therapy-experienced patients, the CD45RO positive (memory) CD4 Tcell (mT4) compartment was most frequently positive (19/21)for HIV-1 proviral DNA, but 6/17 CD45RA positive (naıve)CD4 T (nT4), 5/21 memory CD8 (mT8), and 1/20 naıve CD8(nT8) compartments were also positive for HIV-1 proviralDNA (compartments that positively amplified are identifiedby shading). A required number of cells (i.e., 1 x l04) were notavailable (NA) for four nT4 and one nT8 compartments. Onlyone sample (#31) was positive in three compartments (in-cluding mT4, nT4, and mT8). In the table, T cell sub-compartments that housed CXCR4- and CCR5-binding HIV-1are designated as (X4) and (R5), respectively. It should benoted that three out of four samples that harbored the CXCR4-binding HIV-1 (i.e., samples 13, 14, 26, and 27) had eithernondetectable or <48 HIV-1 RNA copies/ml plasma viralload. Only one sample, #13, had a reasonably high plasmaviral load of 27,094 HIV-1 RNA copies/ml.

In contrast, 14/48 compartments in therapy-naıve patientswere positively amplified for HIV-1 proviral DNA, but theinfection was largely confined to the mT4 subset. Only 2/12therapy-naıve samples that were amplified displayed thepresence of HIV-1 in mT4 and nT4 compartments; these werealso the only two samples containing X4 virus. HIV-1 was notdetected either in mT8 or nT8 among the therapy naıvesamples.

Phylogenetic relationship among the viral lineagesin different T cell subsets

A MEGA neighbor-joining tree was drawn for the HIV-1C2V3 genetic clones of the 11 samples from therapy-experiencedpatients that yielded at least two positive T cell subsets (Fig. 1).It should be noted that the virus in each T cell subcompart-ment formed a separate cluster apart from the virus in theother T cell compartments, except in sample #14. HIV-1 wasamplified in the mT4 (open triangles) and nT4 (open circles)compartments of six samples (#6, 10, 14, 21, 25, and 30) and inmT4 (open triangles) and mT8 (closed triangles) in 4 samples(#16, 18, 22, and 27). One sample (#31) was positive in threecompartments, including nT4, mT4, and mT8. Figure 2showsindividual neighbor joining trees of four examples (samples#6, 16, 22, and 31) with respective evolutionary distancescalculated by Tamura–Nei nucleotide substitutions andbootstrapping values. Another tree was drawn for the twelvesamples from therapy-naıve patients (Fig. 3), all of whichsamples, upon PCR amplification, showed the presence ofHIV in mT4. Two (#18, 21) were also positive in nT4, but noCD8 T cell compartment in any of the samples was HIV-1positive.

Divergence of viral sequences within each compartmentand genetic distances between compartments were calculatedby using the MEGA program, and the results are summarizedin Table 2. Significant separation of the compartments is in-dicated by bold face of p values calculated by the Student’s t-test. Only two samples, #21 (experienced) and 18SD (naıve),were judged to be intermingled, but in all other samples, HIV-1 in different T cell subsets were clearly distinct from eachother as also visually evident. Out of six samples from ther-apy-experienced patients, whose mT4 and nT4 were HIV-1

positive (Fig. 1), the two T cell compartments in samples 6, 10,14, and 25 had the virus, which was quite homogeneous withthe intra- and inter-compartmental genetic distances of0.1*1.8% and 0.7*3.2%, respectively. The virus in the othertwo samples (21 and 30) was somewhat more diverse, withthe respective genetic distances of 0.3%*4.9% and 2.4*5.3%( p¼ 0.12 and 0.48 for intra- and inter-compartmental diver-gence). Sample 35, in which three compartments (mT4, nT4,and mT8) were HIV-1 positive, showed a similar pattern, withthe intra- and inter-compartmental genetic distances ranging0.3*3.5% and 2.8*3.5%, respectively. Overall, intra- or inter-compartmental genetic distances were not very much differ-ent between the therapy-experienced and -naıve groups ofpatients ( p¼ 0.94 and 0.046, respectively).

Of the four samples from therapy-experienced patients inwhich both mT4 and mT8 were positive for HIV-1 (Fig. 1), two(18 and 27) had intra- and inter-compartmental distances of0.3*1.7% and 0.7*1.3%, respectively, in contrast to the othertwo (16 and 22) that showed greater inter-compartmentaldistances (5.8%*14.1%) though the differences were notstatistically significant ( p¼ 0.29 and 0.60 for intra- and inter-compartmental distances). Sample 31, which was positive inthree compartments, seems also to fit the former pattern.Coreceptor binding characteristics of the virus were deter-mined in 44 compartments and included 12 X4 and 32 R5HIV-1 strains (Table 1). Those compartments with X4 virusseemed to remain more homogeneous than those with R5virus (intra-compartmental genetic distances of 0.4% vs. 1.6%,respectively) in both therapy-experienced ( p< 0.0001) and-naıve patients ( p¼ 0.032).

Directional HIV-1 gene flow among differentT cell subsets

A Bayesian consensus tree was created in each case after 3million generations of MCMC analysis, at which point alleffective sample size parameters attained the >100 level asrecommended.26–29 HIV-1 gene flows between the compart-ments, as assessed by using the MacClade program, aresummarized (Table 3) for ART-experienced (top) and -naıve(bottom) patients. Overall, there seemed to be less flowamong the compartments in ART-experienced than in -naıvepatients, though the significance of this could not be discernedbecause of the small sample sizes. In ART-experienced pa-tients, for example, HIV-1 gene flows were assessed betweennT4 and mT4 in six samples (#6, 10, 14, 21, 30, and 31) andbetween mT4 and mT8 in four samples (#16, 22, 27, and 31).Even though a significant (>1.0) gene flow was indicated fromnT4 to mT4 in case #14 (5.9), gene flows were significant in thereverse direction in three cases (#10, 21, and 30). No significantflow existed between mT8 and either CD4 T cell subset. Incontrast, in two ART-naıve cases (#DR18 and DR21), HIV-1flow seemed to be greater from nT4 to mT4 (2.4 and 16.5 vs. 1.4and 3.4, respectively), though the statistical significancecould not be deduced because of the sample size (n¼ 2). Fig-ures 4 and 5a/b show consensus trees of HIV-1 clone se-quences of the involved compartments from a single ART-experienced case #31 (Fig. 4) and two ART-naıve cases (Fig. 5aand b), respectively. In all analyses, 3,000,000 re-samplingsachieved the effective sample size posterior probability scoreof >100, as calculated by the program Tracer (v 1.5) (http://tree.bio.ed.ac.uk/software/tracer/), which was recommended

402 PEREZ ET AL.

by other studies.27–29 Respective bubblegrams also providedvisual comparison of the viral gene flows. The three T cellcompartments (nT4, mT4, and mT8) in case #31 were wellinsulated from each other, with only weak viral gene flows(1.0) from the mT4 to nT4 and mT8 compartments. No re-verse flow was detected. In two ART-naıve cases (Fig. 5a andb), the viral gene flows seem to be in the direction of nT4 to

mT4 (2.4 and 16.5) rather than the reverse (1.4 and 3.36),respectively.

Discussion

The memory CD4 T cell (mT4) subset has long been iden-tified as the major target of HIV-1 infection.10–13 Indeed, in 33

FIG. 1. Neighbor-joining tree (MEGA) of HIV-1 env C2-V3 genetic clones isolated from the purified T cell subsets of elevenART-experienced HIV-1 patients. For each ART-experienced HIV-1 infected patient, peripheral blood mononuclear cells wereisolated by centrifugation on a Ficoll-gradient and naıve and memory subsets of CD4 and CD8 T cells, respectively, wereisolated by cell sorting using a FACSAria. After cell lysis, DNA was purified and amplified for HIV-1 env C2–V3 proviralgene sequence. The amplicons were cloned and approximately 35 genetic clones were picked and sequenced for eachcompartment. Ten clonal sequences were randomly chosen for each compartment and a MEGA neighbor joining tree wascreated (10,000 bootstrapping) to demonstrate the genetic relationship among HIV-1 infecting different T cell subsets.

HIV-1 DISTRIBUTION IN T LYMPHOCYTE SUBSETS 403

patients studied, 31 out of 45 (69%) HIV-1-positive compart-ments were mT4, and among therapy-naıve samples theprevalence rose to 86% (12 of 14). However, other T cellcompartments were also HIV-1 infected, particularly in

therapy-experienced patients. In total, 31 T cell compartmentswere HIV-1 positive in 21 therapy-experienced patients (1.48compartments per patient) as opposed to 14 compartments in12 therapy-naıve patients (1.17 compartments per patient). It

LEGEND:

HXB2 C2V3 CD4RO

CR4RA CD8RO

9999

9292

0.020.02

79

73

0.020.02

95959999

72726868

0.010.01

9898

9595

8585

8181

6969

7171

7171

0.010.01

#6 #16

#22 #31

FIG. 2. Individual neighbor joining trees for four examples; samples #6, 16, 22, and 31. Individual NJ trees for samples #6,16, 22, and 31 are drawn using MEGA program to a scale and bootstrapping values of the analyses are shown in each figure.

404 PEREZ ET AL.

may be argued that the slightly higher number of HIV-1-infected T cell compartments in the former group ( p¼ 0.170;Fisher’s exact test) probably reflected longer infection times(older infections) than that found in the latter. While thiscould not be accurately assessed, the similarity of intra- andinter-compartmental divergence rates in both groups tends toargue against the reasoning. Furthermore, the number of HIV-1-positive T cell compartments was not directly associatedwith the CD4 T cell counts in the compartments of eithergroup. Because the viral loads of the majority of the partici-pants were unknown, it was not possible to determine thepossible influence of viral loads on the number of compart-ments infected. However, as the presence of provirus, par-ticularly in memory subsets, would not require infection bycurrent viral replication, it is unlikely that a strong correlation

would exist between a single viral load measurement andcompartmental distribution.

In therapy-naıve patients, no CD8 T cell compartment wasHIV-1 infected. The numbers of CD8 cells in naıve sampleswere uniformly higher, indicating that the absence of infectionwas not due to lack of the target cells. The virus could bedetected in five mT8 and one nT8 compartment in thetherapy-experienced patients (Table 1). Of the five mT8-positive samples (#16, 18, 22, 27, and 31), which were also HIVpositive in either mT4 or nT4 (or both in #31), three (#16, 18 &22) had R5 virus exclusively, while the other two had X4 in atleast one compartment. Interestingly, sample #31 harboredR5 virus in mT4 but X4 in the nT4 and mT8 compart-ments. Sample #13, the only sample that was positive in thenT8 compartment, also harbored X4 virus. The ratio of

FIG. 3. Neighbor-joining tree (MEGA) of HIV-1 env C2-V3 genetic clones isolated from the purified T cell subsets of twelveART-naıve HIV-1 patients. For each ART-naive HIV-1 patient, C2-V3 genetic clones were isolated as described in the Figure 1legend. Genetic relatedness among virus infecting different T cell subsets was assessed by using MEGA program, also asdescribed in the legend to Fig. 1.

HIV-1 DISTRIBUTION IN T LYMPHOCYTE SUBSETS 405

Table 2. Intra- and Inter-Compartmental Genetic Distances of HIV-1 in Different T Cell Subsets

Intra-compartmental distance

ID number 4RA 4RO 8RA 8ROInter-compartmental

distance p Value (t-test) CD4 count

ART Experienced3 0.045 < 2005 0.01610 0.007 0.017 0.025 < 0.000113 0.00115 0.03726 0.0016 0.003 0.018 0.032 0.032 200* 5007 0.00821 0.049 0.046 0.053 (0.39)24 0.00525 0.001 0.001 0.031 0.000827 0.003 0.003 0.013 0.0119 0.009 > 50020 0.00322 0.023 0.006 0.058 0.01918 0.003 0.017 0.017 < 0.000114 0.005 0.008 0.007 0.009630 0.003 0.025 0.024 < 0.000116 0.008 0.002 0.141 0.00617 0.00931 0.006 0.035 0.003 0.035 (nT4/mT4) <0.0001

0.035 (nT4/mT8) <0.00010.028 (mT4/mT4) <0.0001

(Mean) (0.019) (0.015) (0.001) (0.007)

ART Naive26SD 0.002 200* 50027SD 0.04320SD 0.04018SD 0.012 0.019 (0.15)23SD 0.04821SD 0.020 0.049 0.0056SD 0.004 > 5009SD 0.00619SD 0.04417SD 0.00422SD 0.036(Mean) (0.016) (0.027)

Intra- and inter-compartmental genetic distances were calculated using the MEGA program.Those compartments that harbored X4 virus are identified with bold figures.*The significance of the separation between the compartments was determined by Student’s t-test ( p< 0.05).

Table 3. HIV-1 Gene Flow Analysis (Frequency of Changes Between States) Between Two T cell

Subcompartments in 2 ART-Naıve and 9 ART-Experienced Patients

ART-experienced

From\to nT4 mT4 mT8nT4 — 0; 0; 0.4; 5.9; 0; 0.5 [1.13] 0mT4 1.0, 2.0; 0.4; 0.9; 3.0; 1.5 [1.43] — 1.0; 0.4; 1.0; 0.4 [0.7]mT8 0 0; 0.4; 0; 0.4 [0.2] —

ART-naıve

From\to nT4 mT4nT4 — 2.4; 16.5 [9.45]mT4 1.4; 3.4 [2.49] —

406 PEREZ ET AL.

compartments infected by X4 vs. R5 virus was 8/23 (0.348) inthe ART-experienced group as compared to 4 /10 (0.400) inthe ART-naıve group ( p¼ 0.558). Since X4 virus generallyappears at later stages of HIV-1 infection, a similar X4/R5ratio may also support our assumption that the length of in-fection was not significantly different between the twogroups. A previous study demonstrated that nT4 cells wereprimarily infected by syncytium-inducing (i.e., X4) virus 13.

Our present study suggests that this may only apply to ther-apy-naıve cases where both of the HIV-positive nT4 com-partments harbored X4 virus. In contrast, only two out of sixHIV-1–positive nT4 compartments of the ART-experiencedpatients harbored X4 virus. There are a few previous studiesthat monitored the proviral DNA changes of the coreceptorrequirements during antiretroviral therapies.32 However,none of them to our knowledge examined the differences

FIG. 4. BEAST consensus tree and bubblegram (MacClade) of HIV-1 env C2-V3 clonal sequences isolated from three T cellcompartments of patient #31. A Bayesian consensus tree (top) was created for the HIV-1 C2-V3 clones from three different Tcell subsets by using BEAST program after 3 x 106 resamplings. The clones from nT4, mT4, and mT8 compartments areidentified by solid black, checked block, and shaded blocks, respectively. Directional gene flows of HIV-1 among the infected T cellsubsets were determined by the MacClade program. Numbers in the bubbles represent the directional migration rates andthose over 1.0 are considered to be significant positive flow.

HIV-1 DISTRIBUTION IN T LYMPHOCYTE SUBSETS 407

among the HIV-1 compartments in different T cell subsets.While a previous report 13 characterized the infection of thenT4 compartment by X4 virus as being directly associatedwith the decline of CD4 T cell count, it should also be notedthat both of our therapy-naıve patients who harbored X4 virusin the nT4 compartment had CD 4 T cell counts of more than1,000. At present, we have no reasonable explanation for thisdifference.

Our attempt to determine the direction of viral gene flowamong the T cell subcompartments did not yield conclusiveresults, largely due to the small number of samples that wereHIV-1 positive in more than one compartment. Sample #31,which was positive in 3 T cell compartments including nT4,mT4, and mT8, requires a more careful analysis. While all ofthe other samples contained either X4 or R5 virus exclusivelyin any of their compartments, mT4 harbored R5 virus, whiletwo other (nT4 and mT8) compartments harbored X4 virus.The HIV-1 infected mT8 compartment was previously sug-gested to represent ‘‘prothymic’’ CD4þCD8þ T cells that wereinfected by CD4-binding virus, which then further differen-tiated as CD4-CD8þ T cells.16 Alternatively, the subset mayrepresent activated, cytotoxic CD8 T cells that were HIV-1infected. A number of studies have demonstrated that in vitrocostimulation of peripheral CD8 T cells with anti-CD3 andanti-CD28 monoclonal antibodies induced the expression ofCD4 markers.34–37 When infected by HIV-1 through the newlyexpressed CD4 receptor,38 the receptor becomes againdownregulated as the viral replication begins.39–41. Such cellsmay be phenotypically identical to HIV-1–positive mT8 de-scribed above. It is interesting to note that inter-compart-mental genetic distances were seemingly smaller when two Tcell compartments were infected by X4 virus (e.g., samples#27), which involved mT4 and mT8, and #14, in which nT4and mT4 were infected by X4 virus. In comparison, when

more than one T cell compartment was infected by R5 virus,inter-compartmental genetic distances were significantly lar-ger. Sample #31, in which X4 virus was identified in nT4 andmT8, while R5 virus was associated with the mT4 compart-ment, seemed to follow the pattern associated with R5 virusinfection. Obviously, this observation needs to be further in-vestigated with a larger number of samples. It should, how-ever, be added that intra-compartmental divergence rateswere also smaller when X4 virus was involved than when theinvolved compartments were infected by R5 virus.

The present findings obviously have several limitations.First of all, the PCR detection of HIV-1 cDNA in any of theisolated T cell compartments was limited by our currenttechnology. It is not possible to rule out those other T cellcompartments, which we failed to amplify, that still containeda very low copy number of HIV-1 cDNA. This may be par-ticularly important in samples from ART-naıve patients. Inthe present study, two out of ten ART-naıve samples wereHIV-1 positive in two compartments. It is therefore possiblethat HIV-1 infection in more than one compartment may oc-cur in ART-naıve patients more often than our current dataindicate. However, it is safe to suggest that the HIV-1 viralburden in any T cell compartments other than mT4 is verylight in ART-naıve patients. Our current technique is capableof amplifying and then sequencing 2 cDNA copies of the HIV-1 C2-V3 gene per reaction mixture. We will attempt to furtherconcentrate the possible HIV cDNA template or, alternatively,to increase the number of sorted T cell subset in our futureexperiments. Additionally, PCR amplification followed bycloning that was used in our present study may be technicallybiased for selecting the majority viral populations. Conse-quently, the diversity estimated for each compartment may besignificantly underestimated as previous discussed by Liuet al.42 We tried to partially correct this bias by using a serial

FIG. 5. BEAST consensus trees and bubblegrams (MacClade) of HIV-1 env C2-V3 clonal sequences isolated from the nT4and mT4 compartments of ART-naıve patients DR18 (A) and DR21 (B). Bayesian consensus trees (top) were created for theHIV-1 C2-V3 clones from the two different T cell subsets by using BEAST program after 3�106 resamplings. The clones fromnT4 and mT4 compartments are identified by solid black and checked blocks, respectively. Directional gene flows of HIV-1between the infected T cell subsets were determined by the MacClade program. Numbers in the bubbles represent thedirectional migration rates and those over 1.0 are considered to be significant positive flow.

408 PEREZ ET AL.

dilution of each template and picking amplicons at the lim-iting dilution and at x100 highest concentrationfor clon-ing.20,21 However, because of the small number of clones usedfor analysis, such biases may still remain.

Even though our attempts to determine the direction ofviral gene flow among different T cell subsets did not yieldconclusive evidence, this type of analysis is importantfor determining how HIV-1 migrates among different targetT cell populations. Even though a consistent directionalflow of the virus could not be proven, there was an indi-cation of strong gene flow from nT4 to mT4 in case #14(5.9), while reverse directional flows (i.e., from mT4 to nT4)of 2.0 (case #10) and 3.0 (case #21) were also detected. Viralflows between mT4 and mT8 were negligible in all casesinvolved. We are currently utilizing the presence of drugresistance-associated and synonymous mutations in theviral pol gene as the additional determinants of the direc-tional viral flow among T cell subsets in ART-experiencedpatients, the results of which, however, will be describedelsewhere.

Acknowledgments

The present study was supported in part by RCMI infra-structure Grant G12RR003050 and CCHD Grant U54RR019507, both of which were awarded by the NIH-NCRR.Thanks go to the RCMI Publications Office, to Mr. Bob Ritchiein particular, for editing this manuscript.

Author Disclosure Statement

No competing financial interests exist.

References

1. Bello G, Casado C, Garcia S, et al.: Co-existence of recent andancestral nucleotide sequences in viral quasispecies of hu-man immunodeficiency virus type 1 patients. J Gen Virol2004;85:399–407.

2. Coffin JM: HIV viral dynamics. AIDS 1996;10:S75–84.3. Korber BTM, Kunstman KJ, Patterson BK, et al.: Genetic

differences between blood- and brain-derived viral se-quences from human immunodeficiency virus type 1-infected patients: Evidence of conserved elements in the V3region of the envelope protein of brain-derived sequences. JVirol 1994;68:7467–7481.

4. Cheng–Meyer C, Weiss C, Seto D, and Levy JA: Isolates ofhuman immunodeficiency virus type 1 from the brain mayconstitute a special group of the AIDS virus. Proc Natl AcadSci USA 1989;86:8575–8579.

5. Dittmar MT, Simmons G, Donaldson Y, et al.: Biologicalcharacterization of human immunodeficiency virus type 1clones derived from different organs of an AIDS patient bylong-range PCR. J Virol 1997;71:5140–5147.

6. Tirado G, Jove G, Kumar R, et al.: Compartmentalization ofdrug resistance-associated mutations in a treatment-naiveHIV-infected female. AIDS Res Hum Retroviruses 2004;20:684–686.

7. Tirado G, Jove G, Reyes E, et al.: Differential evolution ofcell-associated virus in blood and genital tract of HIV-infectedfemales undergoing HAART. Virology 2005;334:299–305.

8. Paranjpe S, Craigo J, Patterson B, et al.: Subcompart-mentalization of HIV-1 quasispecies between seminal cellsand seminal plasma indicates their origin in distinct genitaltissues. AIDS Res Hum Retroviruses 2002;18:1271–1280.

9. Philpott S, Burger H, Tsoukas C, et al.: Human immunode-ficiency virus type 1 genomic RNA sequences in the femalegenital tract and blood: compartmentalization and in-trapatient recombination. J Virol 2005;79:353–363.

10. Douek DC, Brenchley JM, Betts MR, et al.: HIV preferentiallyinfects HIV-specific CD4þ T cells. Nature 2002;417:95–98.

11. Blaak H, van’t Wout AB, Brouwer M, et al.: In vivo HIV-1infection of CD45RA(þ) CD4(þ) T cells is established pri-marily by syncytium-inducing variants and correlates withthe rate of CD4(þ) T cell decline. Proc Natl Acad Sci U2000;97:1269–1274.

12. Ribeiro RM, Hazenberg MD, Perelson AS, and DavenportMP: Naıve and memory cell turnover as drivers of CCR5-to-CXCR4 tropism switch in human immunodeficiency virustype 1: Implications for therapy. J Virol 2006;80:802–809.

13. Brenchley JM, Hill BJ, Ambrozak DR, et al.: T cell subsetsthat harbor human immunodeficiency virus (HIV) in vivo:Implications for HIV pathogenesis. J Virol 2004;78:1160–1168.

14. Fischer M, Joos B, Hirschel B, et al.: Cellular viral reboundafter cessation of potent antiretroviral therapy predicted bylevels of multiply spliced HIV-1 RNA encoding nef. J InfectDis 2004;190:1979–1988.

15. Furtado MR, Callaway DS, Phair JP, et al.: Persistence ofHIV-1 transcription in peripheral-blood mononuclear cells inpatients receiving potent antiretroviral therapy. N Eng JMed 1999;340:1614–1622.

16. McBreen S, Imlach S, Shirafuji T, et al.: Infection of theCD45RAþ (naive) subset of peripheral CD8þ lymphocytesby human immunodeficiency virus type 1 in vivo. J Virol2001;75:4091–4102.

17. Kaiser P, Joos B, Niederost B, et al.: Productive human im-munodeficiency virus type 1 infection in peripheral bloodpredominantly takes place in CD4/CD8 double negative Tlymphocytes. J Virol 2007;81:9693–9706.

18. Saha K, Zhang J, Gupta A, et al.: Isolation of primary HIV-1that target CD8þ T lymphocytes using CD8 as a receptor.Nat Med 2001;7:65–72.

19. Saha K, Zhang J, and Zerhouni B: Evidence of productivelyinfected CD8þ T cells in patients with AIDS: implications forHIV-1 pathogenesis. J Acq Immune Defic Syndr 2001;26:199–207.

20. Zhu T, Muthui D, Holte S, et al.: Evidence for human im-munodeficiency virus type 1 replication in vivo in CD14(þ)monocytes and its potential role as a source of virus in pa-tients on highly active antiretroviral therapy. J Virol 2002;76:707–716.

21. Zhu T, Corey L, Hwangbo Y, et al.: Persistence of extraor-dinarily low levels of genetically homogeneous human im-munodeficiency virus type 1 in exposed seronegativeindividuals. J Virol 2003;77:6108–6116.

22. Kuiken C, Leitner T, Foley B, et al.: HIV sequence compen-dium 2009. HIV sequence database, Los Alamos NationalLaboratory, Los Alamos, NM.

23. Jensen MA, MCoetzer M, van’t Wout AB, et al.: A reliablephenotype predictor for human immunodeficiency virustype 1 subtype C based on envelope V3 sequences. J Virol2006;80:4698–4704.

24. Philpott S, Burger H, Tsoukas C, et al.: Human immunode-ficiency virus type 1 genomic RNA sequences in the femalegenital tract and blood: compartmentalization and in-trapatient recombination. J Virol 2005;79:353–363.

25. Salemi M, Burkhardt BR, Gray RR, et al.: Phylodynamics ofHIV-1 in lymphoid and non-lymphoid tissues reveals a

HIV-1 DISTRIBUTION IN T LYMPHOCYTE SUBSETS 409

central role for the thymus in emergence of CXCR4-usingquasispecies. PLoS ONE. 2007;2:e950.

26. Drummond AJ, Ho SYM, Phillips MJ, and Rambaut A: Re-laxed phylogenetics and dating with confidence. PLoS Biol.2006;4:e88.

27. Edwards CTT, Holmes EC, Wilson DJ, et al.: Populationgenetic estimation of the loss of genetic diversity duringhorizontal transmission of HIV-1. BMC Evol Biol 2006;6:28.

28. Thomas GH: Phylogenetic distribution of British birds ofconservation concern. Proc Biol Sci 2008;275:2077–2083.

29. Bello G, Aulicino PC, Ruchansky D, et al.: Phylodynamics ofHIV-1 circulating recombinant forms 12_BF and 38_BF inArgentina and Uruguay. Retrovirology 2010;7:22.

30. Abecasis AB, Lemey P, Vidal N, et al.: Recombination con-founds the early evolutionary history of human immuno-deficiency virus type 1: Subtype G is a circulatingrecombinant form. J Virol 2007;81:8543–8551.

31. Slatkin M and Maddison WP: A cladistic measure of geneflow inferred from the phylogenies of alleles. Genetics1989;123:603–613.

32. Salemi M, Lamers SL, Yu S, et al.: Phylodynamic analysis ofhuman immunodeficiency virus type 1 in brain compart-ments provides a model for the neuropathogenesis of AIDS.J Virol 2005;79:11343–11352.

33. Sullivan YB, Landay AL, Zack JA, et al.: Upregulation ofCD4 on CD8þ T cells: CD4dimCD8bright T cells constitutean activated phenotype of CD8þ T cells. Immunology 2001;103:270–280.

34. Saracino A, Monno L, Cibelli DC, et al.: Co-receptor switchduring HAART is independent of virological success. J MedVirol 2009;81:2036–2044.

35. Yang LP, Riley JL, Carroll RG, et al.: Productive infection ofneonatal CD8þ T lymphocytes by HIV-1. J Exp Med1998;187:1139–1144.

36. Kitchen SG, Korin Y, Roth MD, et al.: Costimulation of naiveCD8(þ) lymphocytes induces CD4 expression and allowshuman immunodeficiency virus type 1 infection. J Virol1998;72:9054–9060.

37. Flamand L, Crowley RW, Lusso P, et al.: Activation of CD8þT lymphocytes through the T cell receptor turns on CD4gene expression: Implications for HIV pathogenesis. ProcNatl Acad Sci USA 1998; 95: 3111–3116.

38. Kichen SG, Jones NR, LaForge S, et al.: CD4 on CD8þ T cellsdirectly enhances effector function and is a target for HIVinfection. Proc Natl Acad Sci USA 2004;101:8727–8732.

39. Lundquist CA, Tobiume M, Zhou J, et al.: Nef-mediateddownregulation of CD4 enhances human immunodeficiencyvirus type 1 replication in primary T lymphocytes. J Virol2002;76:4625–4633.

40. Tanaka M, Ueno T, Ankara T, et al.: Downregulation of CD4is required for maintenance of viral infectivity of HIV-1.Virology 2003;311:316–325.

41. Magadan JG, Perez–Victoria FJ, Sougrat R, et al.: Multi-layered mechanism of CD4 downregulation by HIV-1 Vpuinvolving distinct ER retention and ERAD targeting steps.PLoS Pathog 2010;6:e1000869.

42. Liu SL, Rodriguez AG, Shankarappa R, et al.: HIV quasis-pecies and resampling. Science 1996;273:415–416.

Address correspondence toYasuhiro Yamamura, Ph.D.

Ponce School of MedicineAIDS Research Program

395 Industrial Reparada-2Ponce, PR 00716-2348

E-mail: [email protected]; [email protected]

410 PEREZ ET AL.